Real-Time RRT^* with Signal Temporal Logic Preferences

Recent advances in Rapidly-exploring random tree: A review

“…The real-time performance of the improved RRT in a static scenario is shown in 3.4.1, and even for dynamic scenarios, the improved algorithm still shows its superior real-time performance. For example, STL-RT-RRT∗ [ 192 ] and RT-RRT∗ [ 193 ] can react to moving obstacles in real scenarios (1.1 m/s for moving obstacles and 0.55 m/s for the robot). STL-RT-RRT∗ even achieves 100 % obstacle avoidance success rate in 1000 trials, with a failure chance of less than 0.1 %.…”

“…In Ref. [ 192 ], a car-like robot on a miniature circular runway at a speed of 0.17 m/s successfully avoid the moving obstacles across the runway. In simple dynamic scenarios, the average dynamic obstacle avoidance success rate of the robotic arm is 100 % and the average path generation time is less than 0.1s [ 161 ].…”

“…One type of uncertain environment is the dynamics of the scene, Alexis et al [ 192 ] introduce a signal temporal logic (STL) for real-time dynamic obstacle avoidance, which has the potential for dynamic obstacle avoidance optimization as it expresses through quantitative semantics that a robot should always keep a safe distance from a human or move slowly in narrow passages. The authors compare the performance of STL-RT-RRT∗ and RT-RRT∗ [ 193 ] and show the better performance of STL-RT-RRT∗ in terms of success rate in dynamic obstacle avoidance (no collisions for STL-RT-RRT∗ and 513 collisions for RT-RRT∗ in 1000 trials), the number of stops affected by dynamic obstacle avoidance (4 times for STL-RT-RRT∗ and 286 times for RT-RRT∗ in 1000 trials), and safe distance from dynamic obstacles (e.g., 1.2 m radius from human).…”

“…Thus, while much of the literature describes their work as stable and highly reliable, e.g., no failures in 1000 trials [ 192 ], in practice, even with very low failure rates (Consider hardware, communication failures, etc. ), some robots may fail to achieve full autonomy, especially in safety-related tasks, such as surgical robots [ 206 ].…”